Microstructural Insights into Fast Ion Transport in Solid Electrolytes via Multiscale Modeling

Technical Summary: Microstructural Insights into Fast Ion Transport in Solid Electrolytes via Multiscale Modeling

Problem Statement
The development of high-performance all-solid-state batteries is hindered by an incomplete understanding of the microstructural features that enable fast ion transport in solid electrolytes. While historical efforts have focused on optimizing bulk superionic conductivity, practical solid electrolytes are polycrystalline, containing dense networks of grain boundaries (GBs). Although GB effects in oxide electrolytes are partially understood, the role of GBs in sulfide electrolytes remains unclear due to experimental difficulties in characterizing morphology and separating GB contributions from bulk conduction. Furthermore, modeling ion transport in polycrystals is computationally challenging: it requires bridging atomic-scale hopping mechanisms with continuum-scale percolation, demanding interatomic potentials that accurately describe complex GB interactions and simulation cells large enough to capture structural disorder, which traditional ab initio methods cannot achieve at relevant scales.

Methodology
The authors propose a hierarchical multiscale modeling framework that integrates density-functional theory (DFT), machine-learning potentials (MLPs), molecular dynamics (MD), and finite element (FE) simulations.

1. Machine-Learning Potentials (MLPs): To balance accuracy and computational cost, the authors developed lightweight moment tensor potentials (MTPs) using a closed-loop active learning scheme. This process involved extracting local atomic environments from full microstructures containing GBs, performing DFT calculations on these local structures, and iteratively training MTPs with increasing complexity until near-ab initio accuracy was achieved.
2. Molecular Dynamics (MD): The trained MLPs were employed in large-scale MD simulations (up to nanoseconds) of surrogate GB models, including bicrystals, columnar crystals, and polycrystals with random GB networks. These simulations captured Li-ion disorder, correlations, and diffusion kinetics at temperatures ranging from 300 K to 600 K. Key metrics included self-diffusivities, diffusion barriers, and excess entropy to quantify structural disorder.
3. Finite Element (FE) Simulations: MD-derived self-diffusivities and GB widths were used as inputs for continuum-scale FE simulations. To handle the scale separation, two approaches were utilized:
   • Resolved GBs: For nanograins (~25 nm), GBs were volume-resolved using Fourier-accelerated nodal solvers (FANS).
   • Collapsed GBs: For micrograins (~1 µm), GBs were represented as thickness-collapsed finite elements.
     This approach allowed the prediction of macroscopic diffusivities and conductivities for polycrystalline systems with device-relevant grain sizes.

Key Contributions and Results
The study was applied to argyrodite solid electrolytes Li6PS5X (where X ∈ {Cl, Br, I}), systematically investigating the effects of anion ordering, halide substitution, and GB morphology.

• GB Energetics and Structure: In anion-ordered Li6PS5Cl, GB formation energies were found to be low (12–22 meV Å⁻² for general GBs, ~5 meV Å⁻² for Σ3), suggesting abundant and random GBs in synthesized samples. GBs induce structural disorder, evidenced by increased excess entropy and sub-nanometer voids, though these voids have minimal impact on diffusivity.
• Anion Ordering and Substitution Effects:
  • Anion Disorder: In anion-disordered Li6PS5Cl, GBs act as barriers, reducing Li-ion self-diffusivity by 60% relative to the bulk. However, the characteristic hopping frequencies remain high (10⁸ Hz), consistent with low GB resistance observed experimentally.
  • Halide Substitution: A linear scaling relationship was established between the Li-ion diffusion barrier and the anion radius in ordered phases. Increasing the anion radius (Cl → Br → I) increases the diffusion barrier by ~80 meV per 0.1 Å increase.
  • Opposite GB Effects: The impact of GBs is bulk-dependent. In fast-diffusing phases (anion-disordered Li6PS5Cl), GBs suppress diffusion. In slow-diffusing phases (anion-ordered Li6PS5I), GBs enhance diffusion by up to 60-fold, acting as connecting channels.
• Non-Arrhenius Transport: The simulations successfully reproduced the non-Arrhenius transport behavior observed in anion-ordered Li6PS5I, identifying a transition temperature (~250 K) where the transport mechanism shifts from bulk-dominated (high temperature) to GB-dominated (low temperature). This behavior is attributed to the lower activation energy of GBs compared to the bulk in this specific system.
• Grain-Size Effects: Continuum simulations revealed that reducing grain size to the nanoscale mitigates non-Arrhenius behavior. For Li6PS5Cl, grain refinement improves intergranular contact without compromising superionic conductivity, whereas for Li6PS5I, nanosizing activates transport in systems lacking intrinsic superionic behavior.

Significance and Claims
The paper claims that its multiscale modeling framework provides a rigorous, quantitative link between atomic-scale mechanisms and macroscopic device performance, resolving the long-standing challenge of modeling ion transport in polycrystalline solid electrolytes.

• Mechanistic Insight: The study clarifies that GBs do not universally impede ion transport; rather, their effect (enhancement vs. suppression) is dictated by the intrinsic diffusivity of the bulk phase.
• Design Guidelines: The results offer a priori guidance for microstructural design. For fast-diffusing sulfides like Li6PS5Cl, grain refinement is beneficial for contact without sacrificing conductivity. For slow-diffusing systems like Li6PS5I, nanosizing can be a strategy to activate transport.
• Compositional Parameters: The work identifies the anion radius as a key compositional parameter governing diffusion barriers in ordered argyrodites, challenging the notion that lattice polarizability is the sole dominant factor.
• Experimental Validation: The predicted non-Arrhenius behavior and transition temperatures in Li6PS5I align with experimental observations, suggesting that GBs are the origin of this phenomenon in these materials.

The authors conclude that this approach enables the design of advanced solid electrolytes by explicitly accounting for microstructural complexity, moving beyond idealized bulk models to realistic polycrystalline representations.